AutoGPT（现 Experts AI 团队平台）AutoGPT (now the Experts AI-team platform)
AutoGPT 现役形态是「你的 AI 专家团队」（首页当日招牌：Your team of AI Experts——按职位分工的 AI SDR/SEO writer/LinkedIn 写手/执行助理，由 Otto 协调）：托管平台 platform.agpt.co + 开源自托管双路径， 仓 ★187,826 当日核验；从 2023 自主 agent 实验到低代码平台再到 Experts，产品线两度转向， classic 线仍在仓但已非主打——按任何一代旧印象建档都会错位。
AutoGPT's current form is "your team of AI Experts" (same-day homepage headline: Your team of AI Experts — role-based AI SDR / SEO writer / LinkedIn writer / executive assistant, coordinated by Otto): hosted platform platform.agpt.co + open-source self-hosting, repo ★187,826 verified same day; from the 2023 autonomous-agent experiment to a low-code platform to Experts, the product line pivoted twice — the classic line still lives in the repo but is no longer the lead, and building a profile from any single generation's impression misleads.
适合与不适合When to use it, when not to
适合：多步异步的委托式「AI 团队」工作（内容/销售/研究）、要开源自托管。 不适合：要确定性画布工作流、要大陆直连备案入口、要平台区宽松许可（Polyform Shield）。
Fits: multi-step asynchronous delegated "AI team" work (content/sales/research), open-source self-hosting. Not a fit: deterministic canvas workflows, a filed direct-mainland entry, a permissive platform-zone license (Polyform Shield).
中文生态China availability 国产 / 国内可直接用 / 备案状态 · 快照domestic / direct access / CAC filing · snapshot 2026-10-30
- 出品属地Origin
- 国外出品foreign-made Determinist Ltd（英格兰及威尔士注册，trading as "AutoGPT"；条款与隐私政策当日直抓在册）。Determinist Ltd (registered in England and Wales, trading as "AutoGPT"; on record in the terms and privacy policy direct-fetched same day).
- 国内使用Direct access
- 大陆可直连direct OK agpt.co、/legal/* 条款页与 GitHub 仓大陆直连实测 200（2026-10-30）；platform.agpt.co 托管入口未单独实测。agpt.co, the /legal/* terms pages and the GitHub repo measured 200 direct from mainland (2026-10-30); the platform.agpt.co hosted entry was not separately measured.
- 备案状态CAC filing
- 未见于备案名单not in filing list：未见于网信办公示名单（2026-10-30 检索百炼公示页与第十六批批次清单，未见 AutoGPT/Determinist 条目）；该产品未面向大陆公众提供服务，备案不适用。not in the CAC public registry (Bailian disclosure page and the 16th batch list searched 2026-10-30; no AutoGPT/Determinist entry); the product does not serve the mainland public, so filing does not apply.
备案信息以网信办《生成式人工智能服务已备案信息》公示名单为准（按批次滚动更新，本站只抄编号不解读）；「大陆可直连」是本站从大陆网络直连实测的结果，不是厂商承诺。GitHub 类站点当日实测有波动，口径写在依据里。Filing status is quoted from the CAC public registry (updated batch by batch; this site copies the number and does not interpret it). "Direct access" is this site's own reachability measurement from a mainland network, not a vendor promise. GitHub-hosted sites fluctuated on the measurement day — the note says so.
固定坐标系Fixed coordinate system
8 个维度dimensions, 与同分区其他对象逐项可比each comparable item by item against others in the same section.
- 模型与开放条件Model & openness
平台按任务接入各家模型（现役 Experts 线的模型选择面未逐条核验）；classic 自主 agent 线 历史上以 OpenAI 系为主，现役默认未核验。
The platform integrates various models per task (the current Experts line's model selection not verified item by item); the classic autonomous-agent line historically leaned on OpenAI-family models; the current default not verified.
- 运行位置Where it runs
双形态：托管平台（隐私政策当日在册：「The AutoGPT Platform at platform.agpt.co」）+ 开源自托管（导航 GitHub「Explore the code and self-host」在册；仓当日推送 2026-10-30）。
Two forms: hosted platform (privacy policy on record same day: "The AutoGPT Platform at platform.agpt.co") + open-source self-hosting (nav GitHub "Explore the code and self-host" on record; repo pushed same day 2026-10-30).
- 本地文件Local files
自托管形态数据在本机（代码与部署自管）；托管版数据在 Determinist Ltd 云端 （隐私政策自任 data controller，当日在册）。
In the self-hosted form data lives on your own machines (code and deployment self-managed); hosted-version data lives in Determinist Ltd's cloud (the privacy policy self-identifies as data controller, on record same day).
- 关机后的任务Tasks after shutdown
现役产品是「AI 专家团队」：按职位分工的 Experts（AI SDR/SEO writer/LinkedIn posts/执行助理） 由 Otto 协调（首页当日在册）——多步异步执行是产品主轴；排队与失败恢复细节未逐条核验。
The current product is a "team of AI Experts": role-based Experts (AI SDR / SEO writer / LinkedIn posts / executive assistant) coordinated by Otto (same-day homepage on record) — multi-step asynchronous execution is the product's main axis; queueing and failure-recovery details not verified item by item.
- 工具与扩展Tools & extensions
Integrations 页在册（Apps and tools your experts can use）；平台 blocks/工作流面在文档与仓在册； 具体连接器清单未逐条核验。
An Integrations page is on record (Apps and tools your experts can use); the platform's blocks/workflow surface is on record in docs and the repo; the concrete connector list not verified item by item.
- 上下文与记忆Context & memory
Experts 的任务上下文与协调面（Otto）是产品核心；跨会话持久记忆机制未核验到官方声明，按未知处理。
Experts' task context and the coordination surface (Otto) are the product core; cross-session persistent memory not verified in official statements, treated as unknown.
- 权限与限制Permissions & limits
托管版主体 Determinist Ltd trading as "AutoGPT"（英格兰及威尔士注册，条款与隐私政策当日在册）； subprocessors 页在册；训练用途开关路径未逐条核验。
Hosted-version entity Determinist Ltd trading as "AutoGPT" (registered in England and Wales, on record in the terms and privacy policy same day); a subprocessors page is on record; training-use toggle path not verified item by item.
- 适合什么任务Best for
适合：多步异步的「AI 团队」委托式工作（内容/销售/研究）、要开源自托管（平台 Polyform Shield+ classic MIT 分区）。 不适合：要确定性画布工作流（自主 agent 血统）、要大陆直连备案入口、要宽松无限制的平台许可 （平台区是 Polyform Shield 非 MIT）。
Fits: multi-step asynchronous delegated "AI team" work (content/sales/research), open-source self-hosting (platform zone Polyform Shield + classic zone MIT). Not a fit: deterministic canvas workflows (autonomous-agent lineage), a filed direct-mainland entry, a permissive unrestricted platform license (the platform zone is Polyform Shield, not MIT).
头号误解Top misconception
- 条款主体是 Determinist Ltd trading as "AutoGPT"（英格兰及威尔士）——GitHub org 名 Significant-Gravitas 不是签约主体，引用主体按条款原文The terms entity is Determinist Ltd trading as "AutoGPT" (England and Wales) — the GitHub org name Significant-Gravitas is not the contracting entity; cite the entity as written in the terms
- 许可按目录分区：autogpt_platform 内 Polyform Shield、外 MIT——「AutoGPT 是 MIT」只对 classic 线成立；仓 license=NOASSERTION 正是双许可的结果Licensing splits by directory: inside autogpt_platform = Polyform Shield, outside = MIT — "AutoGPT is MIT" only holds for the classic line; the repo's license=NOASSERTION is precisely the result of the split
- 产品线已两度转向：自主 agent 实验（classic）→低代码平台→现役 Experts AI 团队（首页当日招牌）——按任一旧印象建档都会整体错位；classic 线仍在仓（classic/ 目录）但已非主打The product line pivoted twice: autonomous-agent experiment (classic) → low-code platform → the current Experts AI team (same-day homepage headline) — building a profile from any generation's old impression misleads wholesale; the classic line still lives in the repo (classic/ directory) but is no longer the lead
未知项清单Known unknowns
- 现役 Experts 线接入的模型清单（未逐条核验）
- 跨会话持久记忆机制（未核验到官方声明）
- 训练用途开关路径与默认态（未逐条核验）
- platform.agpt.co 大陆直连可达性（未单独实测）
- The model list behind the current Experts line (not verified item by item)
- Cross-session persistent memory mechanism (no official statement verified)
- Training-use toggle path and default state (not verified item by item)
- Direct-mainland reachability of platform.agpt.co (not separately measured)
证据来源Evidence sources
判断可回到以下一手源复核。本站核验日Every judgement can be rechecked against the primary sources below. Verified on 2026-10-30，内容更新日content updated on 2026-10-30。
| 类型Type | 名称Name | 链接Link |
|---|---|---|
| official | AutoGPT · 官网（Your team of AI Experts/Experts 分工/Otto 协调，2026-10-30 直抓 200）AutoGPT · official site (Your team of AI Experts / role-based Experts / Otto coordination, direct-fetched 200 on 2026-10-30) | https://agpt.co |
| official | AutoGPT · Platform Terms of Use（主体 Determinist Ltd，2026-10-30 直抓 200）AutoGPT · Platform Terms of Use (entity Determinist Ltd, direct-fetched 200 on 2026-10-30) | https://agpt.co/legal/platform-terms-of-use/ |
| official | AutoGPT · Platform Privacy Policy（platform.agpt.co/Determinist Ltd 自任 controller，版本 2025-04-18，2026-10-30 直抓 200）AutoGPT · Platform Privacy Policy (platform.agpt.co / Determinist Ltd as controller, version 2025-04-18, direct-fetched 200 on 2026-10-30) | https://agpt.co/legal/platform-privacy-policy/ |
| repo | Significant-Gravitas/AutoGPT · 官方仓（GitHub API 当日核验 ★187,826/pushed 2026-10-30；LICENSE 双许可分区当日直抓）Significant-Gravitas/AutoGPT · official repo (GitHub API verified same day: ★187,826 / pushed 2026-10-30; the two-zone LICENSE direct-fetched same day) | https://github.com/Significant-Gravitas/AutoGPT |
| changelog | AutoGPT · GitHub Releases（autogpt-platform-beta-v0.10.0 2026-10-27，作变更记录）AutoGPT · GitHub Releases (autogpt-platform-beta-v0.10.0 2026-10-27, serving as the change log) | https://github.com/Significant-Gravitas/AutoGPT/releases |
实测记录Measured runs
本档未跑工作流搭建实测；结论只覆盖平台能力与权限边界。
实测建议（三条，都指向官方可读页没给的东西）：
- Experts 的任务成功率与人工接管面（委托式 agent 的可靠性核心面，未在册）
- 自托管平台与托管版的功能差（自托管路径在册，两形态差异未在册）
- Otto 协调层的失败恢复与成本上限控制（多 agent 协作的失控风险面，未在册）
This site has not completed measured runs. The coding-oriented test protocol is at protocols/coding/_protocol.md (the coding task suite does not apply to agent platforms).
Suggested measurements (three, all pointing at things the official readable pages do not give):
- Experts' task success rate and human-takeover surface (the reliability core of delegated agents, not on record)
- Feature gap between the self-hosted platform and the hosted version (the self-host path is on record; the delta is not)
- Failure recovery and cost-cap control in the Otto coordination layer (runaway risk of multi-agent collaboration, not on record)
相关条目Related entries
- DifyDify画布式 agent 平台对照：Dify 确定性编排；AutoGPT 自主/委托式血统。Canvas-style agent platform contrast: Dify orchestrates deterministically; AutoGPT carries the autonomous/delegated lineage.
- Relevance AIRelevance AI同为「AI 团队/员工」包装：Relevance 无代码工具集向；AutoGPT 开源+自主 agent 血统。Also packaged as an "AI team/employees": Relevance AI leans no-code toolsets; AutoGPT is open source with an autonomous-agent lineage.
- LangChainLangChain代码优先对照：要源码级控制归 framework，要托管团队归本类。The code-first contrast: source-level control belongs to the framework class, the hosted team to this class.
本页由This page is generated from the ai-handbook 数据层生成（CC BY 4.0）。data layer (CC BY 4.0).
方法论与坐标系定义见仓库内For the methodology and coordinate definitions, see METHODOLOGY.md.